David Warde-Farley
David Warde-Farley
@bartvm and @janchorowski should have a look at this. I think there are probably good reasons to structure it as it is but I don't know what they are off-hand....
io.open https://stackoverflow.com/questions/10971033/backporting-python-3-openencoding-utf-8-to-python-2 On Fri, Apr 29, 2016 at 10:56 AM, Dzmitry Bahdanau wrote: > > It does not seem to work with Python 2: > > Processing /home/travis/build/mila-udem/fuel > >...
Seems like we should change TestServer to use one of the procedurally generated datasets, as that isn't tied to a specific dataset anyway.
Is MultiProcessing actually useful in light of the zmq client-server architecture?
A few comments: - I kind of hate the `has_bias`/`use_bias` duplication and it would be great if we could find a way to reconcile the two into one thing. It...
There are added benefits for refactoring: if we make it such that user imports are always done from the `blocks.bricks`, `blocks.initialization` etc. levels, we are able to move stuff around...
Er, you mean GitHub rather than Travis? Don't think Scrutinizer reports to Travis.
Well that's useful.
I don't think this is practical at all, since AFAIK you cannot have imported Theano in the parent before doing this, and most files in Blocks import Theano. It's an...
> single MongoDB server in the lab ಠ_ಠ Given the way everything else regarding lab infrastructure is run, I think that would probably be quite undesirable. And no matter how...